Deep Vision Transformer with Tasmanian Devil Optimization for Multiclass Paddy Disease Detection and Classification for Precision Agriculture

Abstract

Rice is the daily consumed crop all over the country and other parts of the world. Rice is cultivated in most of the states. Nevertheless, rice plant diseases deteriorate the quantity and quality of the crop. Rice plants are affected by various conditions, for example: sheath blight, foot rot, and so on, producing a loss in the farming yield. Therefore, earlier disease recognition in crops is important. Performing intelligent Farming is a hot zone of investigation to prevent more harm to crops. The extensive growth of Deep Learning (DL) makes it probable to attain the objective of disease recognition in crops. In this manuscript, we introduce a new Deep Vision Transformer with Tasmanian Devil Optimization for Multiclass Paddy Disease Detection and Classification (DViTTDO-MPDDC) technique for Precision Agriculture. The major intention of the DViTTDO-MPDDC technique focuses on the automatic classification and recognition of paddy plant diseases. To accomplish this, the DViTTDO-MPDDC technique uses the wiener filter (WF) technique for the noise removal process. Besides, the vision transformer (ViT) technique is used for feature extraction purposes. Additionally, the attention mechanism-based convolutional neural network with bidirectional long short-term memory (AM-CNN-BiLSTM) technique is used for the paddy disease detection process. Eventually, the TDO algorithm is exploited for the hyperparameter fine-tuning of the AM-CNN-BiLSTM model. To demonstrate the good classification outcome of the DViTTDO-MPDDC algorithm, a wide variety of models occurs on the benchmark database. The extensive comparable findings ensured the betterment of the DViTTDO-MPDDC method over the current methods.